{"id":"e9a15618-78c9-4bbf-8a01-24826b1025e3","arxiv_id":"2508.20929","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"The ATLAS search finds no evidence of heavy neutral leptons in W boson decays and sets the strongest limits to date on their mixing with electron and muon neutrinos in the 15-30 GeV mass range.","lead":"Using 140 inverse femtobarns of LHC proton collisions, the ATLAS experiment searched for heavy neutral leptons, hypothetical particles that could explain why neutrinos have mass. No signal was found, so the collaboration set new upper limits on how strongly these particles mix with muon and electron neutrinos.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified","rationale":"The reader's verdict is ACCEPT, and I agree. The central claim—no excess and exclusion of |U_e|^2>8e-5 and |U_mu|^2>5e-5 over 8-65 GeV—is supported by the presented SR yields, background estimates, validation regions, and CLs limits. The weakest-assumption candidate is the lifetime-reweighting for low-mass signal samples, but on inspection this is not load-bearing: the cτ values corresponding to the quoted excluded mixings at 8-15 GeV are orders of magnitude smaller than 0.1 mm, so the relevant signal is in the prompt regime where the directly generated 0.1 mm sample provides a conservative lower bound on efficiency. Any bias in the reweighting over 0.1-1.0 mm affects only lower, non-excluded mixing values and cannot invalidate the exclusion. The remaining possible concerns (fast simulation, matrix-method closure, asymptotic CLs) are standard, covered by systematic uncertainties, and validated where possible. I therefore recommend no change to the reader's verdict. The suggested concrete test is a useful but non-critical robustness check on the low-mass interpolation/extrapolation boundary.","tokens_in":56502,"tokens_out":40510,"duration_ms":423285,"concrete_test":"Still worth running: for the m=8 GeV point, recompute the 95% CL upper limit on |U|^2 using only the directly simulated cτ=0.1 mm signal efficiency as a fixed proxy for all shorter cτ. If the resulting limit is weaker (higher) than the quoted full-range value, the published limit is conservative and the exclusion claim stands unchanged. If it is stronger (lower), the paper should explicitly document how the sub-0.1 mm efficiency was obtained, since that regime is not covered by the described reweighting.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No load-bearing flaw found in the central exclusion claim. The reader's flagged concern about lifetime reweighting at low mass (Section 3) does not threaten the headline limits: the excluded mixing values at m=8-15 GeV correspond to cτ of order 1e-4 mm (using τ ∝ m_N^-5 |U|^-2), far shorter than the smallest simulated cτ=0.1 mm. Since shorter lifetimes give higher prompt-lepton efficiency, any use of the 0.1 mm sample as a proxy for those shorter lifetimes is conservative; it can only weaken, not strengthen, the derived limit. The reweighting between 0.1 and 1.0 mm is therefore not in the excluded regime, and the 18% uncertainty on it does not drive the result. The dominant background estimate (fake/non-prompt leptons via the matrix method) is validated in nine VRs within 2σ and carries 28-43% systematic uncertainties. The remaining limitations are standard for a collider search: fast simulation for signal and non-public data/software, neither of which constitutes an internal inconsistency.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a search for heavy neutral leptons (HNLs) produced in W-boson decays using the full ATLAS Run 2 dataset of 140 fb^-1 of 13 TeV pp collisions. The search targets prompt, lepton-number-violating signatures: final states with two same-charge same-flavour leptons or three leptons, vetoing three-lepton same-flavour topologies. Backgrounds are estimated through a combination of MC simulation for prompt SM processes and data-driven methods for fake/non-prompt leptons and electron charge-flip, with validation in nine dedicated validation regions. No significant excess is observed; the largest local excess is 1.7 sigma in signal region SRE3. The results are interpreted as 95% CL limits on the HNL mixing parameters |U_e|^2 and |U_mu|^2 for HNL masses 8-65 GeV. The paper claims to exclude |U_e|^2 > 8e-5 and |U_mu|^2 > 5.0e-5 in the full mass range, with strongest limits of |U_e|^2 < 1.1e-5 and |U_mu|^2 < 5e-6 at masses around 15-30 GeV.","tokens_in":56792,"tokens_out":30277,"duration_ms":275029,"significance":"If the limits are correct, this search provides a substantial improvement over the previous ATLAS prompt search based on a partial Run 2 dataset and is competitive with the recent CMS prompt search, placing the strongest prompt-signature constraints on |U_e|^2 and |U_mu|^2 for HNL masses in the 15-30 GeV region. The analysis is technically thorough: the background decomposition is detailed, the data-driven matrix method is validated in nine VRs with agreement within 2 sigma, the systematic uncertainties are propagated through a full profile-likelihood fit with correlated nuisance parameters, and the CLs procedure is used for limits. The paper is clearly written and the internal consistency is strong. The main limitations are the use of fast simulation for signal and the lifetime reweighting for low-mass signal samples, both of which are addressed with assigned systematic uncertainties and appear conservative where the simulation does not directly cover the parameter space.","major_comments":[],"minor_comments":[{"comment":"The sentence 'A uniform 18% uncertainty is applied to the low-mass HNL signal points to take account of uncertainties in the lifetime reweighting technique discussed in Section 8' refers to the lifetime reweighting described in Section 3, not Section 8. Please correct the cross-reference.","section":"Section 7"},{"comment":"The abstract and conclusion state that the strongest muon limit |U_mu|^2 < 5e-6 applies in the mass range 15-30 GeV, while Section 8.3 specifies 20-30 GeV. Please align these statements, or clarify whether the limit also holds at 15 GeV.","section":"Abstract / Conclusion / Section 8.3"},{"comment":"The description of the low-mass lifetime reweighting is terse. It would be helpful to state explicitly how the signal efficiency is obtained for ctau values below 0.1 mm (presumably the 0.1 mm sample is used, which is conservative) and for values between 0.1 and 1 mm (reweighted samples). The sentence in Section 8.3 'For each nominal and intermediate HNL decay length, a limit on the mixing parameter is derived' is ambiguous; clarify that the limit is extracted by scanning |U|^2 and interpolating the signal efficiency as a function of ctau.","section":"Section 3 / Section 8.3"},{"comment":"The claim of exclusion in the full continuous mass range 8-65 GeV is based on limits at discrete mass points (8, 10, 15, 20, 30, 40, 50, 60, 65 GeV). Please state how the limits are interpolated between these points and confirm that the quoted full-range bound is the maximum limit across the tested masses.","section":"Section 8.3 / Figure 4"}],"recommendation":"minor_revision","confidential_remarks":"The stress-test note's estimate that the excluded low-mass HNL points correspond to ctau ~1e-4 mm appears to be too small by orders of magnitude. Using the full HNL width, the relevant ctau values at the m=8-15 GeV limits are approximately 0.1-0.3 mm, which lie within the simulated/reweighted 0.1-1 mm range, so the lifetime-reweighting concern does not threaten the central result. The needed changes are local clarifications and small corrections rather than technical re-analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a textbook ATLAS search paper—well executed, clear, and the central claim (no excess, new limits) holds up. Nothing revolutionary, but it does what it sets out to do.\n\nWhat's new: full 140 fb^-1 Run 2 dataset, extended mass range 8–65 GeV, more signal regions (including two-lepton SRs that recover events with a third baseline lepton), and supersedes the partial Run 2 search [23]. Strongest limits improve on previous results in the 15–30 GeV region.\n\nStrengths: background estimation is careful. The matrix method for fake/non-prompt leptons is validated in nine VRs within 2 sigma, and systematic uncertainties are handled conservatively (28–43% on that dominant background). The likelihood treatment is standard and complete. The paper is honest about its limitations.\n\nSoft spots: the lifetime reweighting for low-mass (8–15 GeV) signal is done with only two generated ctau values (0.1 and 1 mm) and a uniform 18% uncertainty. That's a legitimate modeling choice; if anything it's conservative, since the excluded region corresponds to shorter lifetimes where the 0.1 mm sample is a reasonable proxy. Not a flaw worth blocking. Also, the use of fast simulation for signal and the lack of public data/software are standard for ATLAS analyses—not internal inconsistencies.\n\nThe only mild tension: SRE3 shows a 1.7 sigma excess, which is within background fluctuation. The paper reports it transparently. Citation pattern is appropriate; they cite prior ATLAS, CMS, LHCb, and DELPHI results. No self-citation inflation.\n\nFor whom: anyone working on HNL searches or seesaw phenomenology will want this for the updated limits. It's an incremental but useful result.\n\nRecommendation: yes, send to peer review. It deserves a serious referee. I'd expect acceptance after minor revisions—mostly requests for a few more validation plots or supplementary material, not changes to the physics.","headline":"Clean, incremental ATLAS search: full Run 2 prompt HNL limits, no new technique, no obvious flaws; deserves normal refereeing.","tokens_in":57182,"tokens_out":1610,"would_cite":true,"duration_ms":16407,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Using the full LHC Run 2 data set recorded by ATLAS, the search finds no heavy neutral leptons of mass 8-65 GeV with electron- or muon-neutrino mixing above |U_e|^2 = 8e-5 or |U_mu|^2 = 5e-5, with strongest limits near 1e-5 at 15-30 GeV.","keywords":["heavy neutral leptons","Majorana neutrinos","lepton number violation","same-charge lepton pairs","W boson decay","LHC Run 2","95% confidence limits","prompt signatures"],"falsifier":"Generate dedicated HNL signal events at an intermediate proper decay length (e.g., ctau = 0.3 mm) for masses 8-15 GeV and reconstruct them with the same full detector simulation as the ctau = 0.1 mm sample; a reconstructed efficiency differing by more than 18% from the reweighted prediction would invalidate the low-mass limits. Alternatively, a persistent excess above the predicted background in the combined signal regions, currently 44 observed versus 30 +/- 5 expected, would contradict the null claim.","tokens_in":56460,"feed_emoji":"⚛️","tokens_out":7740,"duration_ms":73834,"temperature":0.7,"pith_summary":"The paper searches for heavy neutral leptons (HNLs, right-handed neutrino partners predicted in seesaw and leptogenesis models) with masses between 8 and 65 GeV, produced in W boson decays in 140 fb^-1 of 13 TeV proton-proton collisions. It looks for prompt decays into three charged leptons plus a neutrino, requiring a same-flavour same-charge lepton pair to exploit the Majorana nature of the HNL and suppress Standard Model backgrounds. No significant excess over background is observed. The analysis therefore excludes electron- and muon-neutrino mixing strengths above 8e-5 and 5e-5 respectively across the full mass range, and reaches its best exclusions of about 1.1e-5 (electron) and 5e-6 (muon) in the 15-30 GeV window. If correct, this closes the prompt-decay gap between earlier partial-run and displaced-vertex searches for HNLs below the electroweak scale.","feed_headline":"W-decay data rule out heavy neutrinos down to 1e-5 mixing","feed_subtitle":"ATLAS full Run 2 search excludes electron- and muon-mixing heavy neutrinos in the 15-30 GeV window at 95% confidence","key_machinery":"The signal-defining object is the same-flavour same-charge (SFSC) lepton pair, which is nearly background-free in the Standard Model and directly encodes the lepton-number-violating Majorana decay of the HNL. The analysis optimises nine signal regions around kinematic variables built from the leptons and missing transverse momentum, including a W-boson mass constraint (the m_test discriminant) that separates signal from top-quark and diboson background. Fake and non-prompt leptons, the dominant background, are estimated with a data-driven matrix method; signal efficiencies for HNL masses below 15 GeV are obtained by reweighting lifetime distributions from Monte Carlo samples generated at cta","core_discovery":"The central claim, stated in the paper's own terms, is that no heavy neutral lepton mixing with electron or muon neutrinos at a strength above the quoted bounds exists in the 8-65 GeV mass range. The search uses a simplified single-HNL model with lepton number violation: W -> l N, N -> l' l'' nu, giving three charged leptons, with the same-flavour pair carrying the same charge. The observed yields in nine signal regions are consistent with backgrounds, with the largest deviation a 1.7 sigma excess in one electron region. Limits derived with the CLs method exclude |U_e|^2 > 8e-5 and |U_mu|^2 > 5e-5 over the full mass range, with the strongest constraints |U_e|^2 < 1.1e-5 and |U_mu|^2 < 5e-6 f","pith_inferences":["The 44 observed events versus 30 +/- 5 expected, with the largest excess in the electron-channel signal region SRE3 (1.7 sigma), suggests a slight background underprediction; if the fake-lepton matrix method misses a component, the electron-channel limits would weaken more than the muon-channel limits.","The low-mass exclusions (8-15 GeV) rest on only two simulated decay lengths; a dedicated full-simulation sample at an intermediate ctau (around 0.3 mm) would directly test the 18% reweighting uncertainty and could sharpen the low-mass boundary.","The explicit veto of three-lepton same-flavour topologies leaves tau-neutrino mixing unexplored; a dedicated tau-flavour search would be the natural complement to this strategy.","With future larger data sets, the same same-charge-lepton selection could push sensitivity toward |U|^2 ~ 1e-6 in the 10-40 GeV window, provided trigger efficiency and fake-lepton systematics keep pace."],"forward_implications":["The full-mass-range exclusions supersede the previous ATLAS prompt search on partial Run 2 data and extend prompt coverage into the 20-65 GeV region.","Combined with displaced-vertex and long-lived searches, no HNL with masses 8-65 GeV and mixings above roughly 1e-5 remains allowed in either prompt or long-lived channels.","Within the single-HNL simplified model, the result directly constrains the parameter space used for leptogenesis and sterile-neutrino dark matter scenarios with sub-electroweak HNL masses.","The limits can be reinterpreted in realistic multi-HNL oscillation models by translating the single-flavour mixing bounds into effective flavour-mixing combinations, as the paper notes."],"supporting_citations":[{"why":"Previous ATLAS prompt HNL search on a partial Run 2 sample that this analysis supersedes and extends to the full data set.","marker":"[23]"},{"why":"ATLAS long-lived HNL search whose displaced-signature coverage is complemented by this prompt search for the 20-65 GeV mass range.","marker":"[25]"},{"why":"CMS full Run 2 prompt HNL search, the recent comparable result this analysis is set against.","marker":"[28]"},{"why":"DELPHI search at LEP1 providing the strongest direct constraints in the 2-75 GeV mass range before the LHC searches.","marker":"[29]"},{"why":"Methodology for reweighting lifetime distributions, used to compute signal efficiencies at intermediate decay lengths for low-mass HNL samples.","marker":"[72]"},{"why":"Measured W boson production cross section times leptonic branching fraction, used to normalise the signal expectation.","marker":"[75]"},{"why":"Formulas for HNL decay width and lifetime as functions of mass and mixing, used to connect mixing strengths to signal efficiencies and limits.","marker":"[73, 74]"},{"why":"Tools and data-driven matrix method used to estimate fake and non-prompt lepton backgrounds, the dominant background in the signal regions.","marker":"[101]"},{"why":"Earlier same-sign/trilepton analysis whose matrix-method application and systematic prescriptions are adopted here.","marker":"[103]"}],"fun_headline_variants":["ATLAS W-decay search excludes heavy neutrinos down to 1e-5 mixing","No heavy neutrinos in W decays: ATLAS sets 1e-5 mixing limit","ATLAS search for heavy neutrinos in W decays: null result, tight limits","W boson decays probe heavy neutral leptons: no excess, strong bounds","Heavy neutral leptons in W decays: ATLAS sets tight mixing limits"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The low-mass limits assume that signal efficiencies for intermediate HNL lifetimes can be obtained by reweighting Monte Carlo samples generated at only two proper decay lengths (0.1 mm and 1 mm), with a uniform 18% uncertainty covering the interpolation; if that reweighting is inaccurate, the 8-15 GeV exclusions would shift.","fun_headline_variants_meta":{"raw":{"variants":["ATLAS W-decay search excludes heavy neutrinos down to 1e-5 mixing","No heavy neutrinos in W decays: ATLAS sets 1e-5 mixing limit","ATLAS search for heavy neutrinos in W decays: null result, tight limits","W boson decays probe heavy neutral leptons: no excess, strong bounds","Heavy neutral leptons in W decays: ATLAS sets tight mixing limits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00101,"raw_usage":{"total_tokens":4170,"prompt_tokens":877,"completion_tokens":3293,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":621,"completion_tokens_details":{"reasoning_tokens":3185}},"tokens_in":621,"tokens_out":3293,"duration_ms":21467,"temperature":1.0,"reasoning_tokens":3185,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T14:41:03.413597+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Generate dedicated HNL signal events at an intermediate proper decay length (e.g., ctau = 0.3 mm) for masses 8-15 GeV and reconstruct them with the same full detector simulation as the ctau = 0.1 mm sample; a reconstructed efficiency differing by more than 18% from the reweighted prediction would invalidate the low-mass limits. Alternatively, a persistent excess above the predicted background in the combined signal regions, currently 44 observed versus 30 +/- 5 expected, would contradict the null claim.","supporting_citations":[{"cited_title":"Search for pair production of squarks or gluinos decaying via sleptons or weak bosons in final states with two same-sign or three leptons with the ATLAS detector","cited_arxiv_id":"2307.01094","evidence_quote":"Earlier same-sign/trilepton analysis whose matrix-method application and systematic prescriptions are adopted here."}],"review_version":1}